A tale of what poor train regulation can lead to...

[whole diary]

[IndustryInsider]

Here's an example of what can happen when trains are regulated badly by Network Rail on the Cotswold Line.

I was at Worcester Shrub Hill tonight, intent on boarding the 17:26 to Oxford. As it was the 16:30 from Hereford, I was watching it on the Live Departures Board on my phone and saw it was 15 minutes late leaving Hereford. That became 20 minutes late at Great Malvern and I started to get that sinking feeling that I was in for a long journey.

The automatic announcements at Shrub Hill were frequent, but contradictory. First it was 11 minutes late, then 7, then 9, 14, 17, and back to 11. All of these were announced and anyone waiting must have wondered whether anybody had a clue when it would actually arrive. A similar spate of announcements greeted passengers waiting the 17:33 to Birmingham!

Anyway we eventually left Shrub Hill at 17:41, 15 minutes late. Not too bad as a bit of padding had allowed the train to make up 5 minutes of the delay. To my dismay we then slowed to a stop at Norton Junction at 17:45. The decision had been made to allow the 17:49 to leave Evesham bound for Shrub Hill in front of us. What a silly decision. We sat at Norton for 20 minutes waiting for the 17:49 to leave Evesham and traverse the single line.

A clear run through to Moreton-In-Marsh followed and we arrived there to see the 18:28 Malvern service sat in the down platform. It was delayed by 20 minutes waiting for us. I then looked at my timetable and got that sinking feeling again - the next down train would leave Oxford at 18:50 and as we would only be at Kingham by then, I was fairly confident we would get stopped again at Ascott. This duly happened and we sat there for 15 minutes looking out into the blackness. Eventual arrival time at Oxford was 19:34. That's 55 minutes late and a total journey time of just under 2 hours from Worcester to Oxford.

So, as a result, we were 55 minutes late. The 16:42 Paddington-Malvern was 20 minutes late, and the 18:25 Worcester-Paddington was delayed 15 minutes at Evesham as a result. There may be further knock-on delays that I am not aware of.

Now imagine what would have happened if we hadn't been stopped at Norton Junction: We left Shrub Hill 15 minutes late, with a clear run and a little padding, we would probably have arrived at Oxford 10 minutes late.

The 15:42 Paddington-Shrub Hill would have waited for us at Evesham and would have been delayed by 15 minutes (at the most). That would have probably meant its return working, the 18:25 from Worcester Shrub Hill to London, would have left about 10 minutes late by the time the crew had changed ends. That would have then probably put 10 minutes of a delay into the 16:42 Paddington-Malvern service.

Compare the delays:
What happened was 55+0+20+15 = 90 minutes
What should have happened: 10+15+10+10 = 45 minutes

Until those responsible have the foresight to look ahead, silly added delays like this will continue to happen. Perhaps there needs to be more staff to allow controllers the time to research the best option, perhaps those that do the job are not up to it? All I know is that I'm going to need most of the bottle of Chablis I've just opened to calm me down!  Rant over.

[Btline]

The automatic announcements at Shrub Hill were frequent, but contradictory. First it was 11 minutes late, then 7, then 9, 14, 17, and back to 11. All of these were announced and anyone waiting must have wondered whether anybody had a clue when it would actually arrive. A similar spate of announcements greeted passengers waiting the 17:33 to Birmingham!

This is normal for Digital Doris and the CISs^▸ (Customer Information System) at Worcester. At least she did not tell you the train would be departing from platform 3 (a south facing bay at WSH)!

[eightf48544]

Compare the delays:
What happened was 55+0+20+15 = 90 minutes
What should have happened: 10+15+10+10 = 45 minutes

Very interesting, I think your table above highlights the problems with the way lateness is currently measured.

In my days on the Southern we added all the delay minutes and divided by the number of trains to give average lateness as well as the % of trains late. I think the threshold was 3 late.

Thus in your example what happened would have given an average dealy of 22.5 minutes. Whereas what could have happened is an average delay of 11.25.

Nowadays the measure seems to be percentage of trains over 5 late compared with total trains run. Thus what happened is best because only 3 out of 4 trains were over 5 late or 25% on time whereas, what could have happened is 4 out of 4 trains 5 late or 0% on time.

Is my assumption correct?

[IndustryInsider]

Nowadays the measure seems to be percentage of trains over 5 late compared with total trains run. Thus what happened is best because only 3 out of 4 trains were over 5 late or 25% on time whereas, what could have happened is 4 out of 4 trains 5 late or 0% on time.

Is my assumption correct?

Pretty much. Ask any commuter whether, over a two week period for example, they'd rather be on time three days, and six minutes late the other seven days. Or, on time seven days and an hour late on three occasions and they'd all say the former. However as the PPM^‡(resolve) is measured the figures would be recorded as 30% and 70% respectively.

I've always wondered whether a better performance measuring system would be based on the average delay at every given station on the trains route. It would dissuade TOC^▸ (Train Operating Company)'s from just adding loads of minutes between the penultimate and final stop of the train, and encourage NR^» (Network Rail - home page) to look at the wider picture. With about 70% of stations now having automatic timing reports for each train generated, and many more getting manual reports from the associated signal box - this may not be as difficult as people think?

[eightf48544]

Thanks for confirming my suspistions.

It used to be very annoying when in the Down Peak from Padd if the semi for Taplow was late for any reason you'd be held to leave behind the next stopper to Slough (pre Dec 2004!) and thus be around 20/25 late at Slough after say a 7 late at Padd. It always struck me then that it was the PPM^‡(resolve)  at fault. As you had one trian on time and one very late  50% punctual rather than 2 late but only by a few minutes each at 0% punctual.

Given the padding in the timetable, even then, if the semi was given a clear run it could have knocked a couple of minutes off by Slough and the stopper would leave a couple of minutes down. It is likely both would be under 5 minutes before the end of their journies and thus 100% punctual.

As poster elesewhere padding between teh last stations leads to sloppy working.

I also agree with your comments that a few minutes down on more journies is better than an hour down on fewer journies.